Two-time state formalism for quantum eavesdropping

The key piece of knowledge which quantum eavesdroppers can access is the correlation between prior and future events, i.e., the postselected results of the legitimate preparations and measurements. We present a method for optimizing eavesdropping strategies which is closely related to the two-time state formalism, the natural way to analyze such scenarios; it converts the task of optimization into an eigenvalue problem. Our framework is applied to the familiar BB84 and B92 protocols as well as to the largely unexplored three-state scheme, which has a remarkable feature: the best eavesdropping strategy does not extract any information about the sent state.